Delta Robot Control Using Single Neuron PID Algorithms Based on Recurrent Fuzzy Neural Network Identifiers

Parallel robot control is a topic that many researchers are still developing. This paper presents an application of single neuron PID controllers based on recurrent fuzzy neural network identifiers, to control the trajectory tracking for a 3-DOF Delta robot. Each robot arm needs a controller and an identifier. The proposed controller is the PID organized as a linear neuron, that the neuron’s weights corresponding to Kp, Kd and Ki of the PID can be updated online during control process. That training algorithm needs an information on the object's sensitivity, called Jacobian information. The proposed identifier is a recurrent fuzzy neural network using to estimate the Jacobian information for updating the weights of the PID neuron. Simulation results on MATLAB/ Simulink show that the response of the proposed algorithm is better than using traditional PID controllers, with the setting time is about 0.3 ± 0.1 (s) and the steady-state error is eliminated. 